Standard benchmarks fall short for multi-turn AI agents. A new approach is the 'job interview eval,' where an agent is given an underspecified problem. It is then graded not just on the solution, but on its ability to ask clarifying questions and handle changing requirements, mimicking how a human developer is evaluated.

AI models show impressive performance on evaluation benchmarks but underwhelm in real-world applications. This gap exists because researchers, focused on evals, create reinforcement learning (RL) environments that mirror test tasks. This leads to narrow intelligence that doesn't generalize, a form of human-driven reward hacking.

Static benchmarks are easily gamed. Dynamic environments like the game Diplomacy force models to negotiate, strategize, and even lie, offering a richer, more realistic evaluation of their capabilities beyond pure performance metrics like reasoning or coding.

There's a significant gap between AI performance in simulated benchmarks and in the real world. Despite scoring highly on evaluations, AIs in real deployments make "silly mistakes that no human would ever dream of doing," suggesting that current benchmarks don't capture the messiness and unpredictability of reality.

Standard AI benchmarks are an engineering tool for measuring performance. A more scientific approach, borrowed from cognitive psychology, uses targeted experiments. By designing problems where specific patterns of success and failure are diagnostic, researchers can uncover the underlying mechanisms and principles of an AI system, yielding deeper insights than a simple score.

OpenAI's evals team is looking beyond current benchmarks that test self-contained, hour-long tasks. They are calling for new evaluations that measure performance on problems that would take top engineers weeks or months to solve, such as creating entire products end-to-end. This signals a major increase in the complexity and ambition expected from future AI benchmarks.

As benchmarks become standard, AI labs optimize models to excel at them, leading to score inflation without necessarily improving generalized intelligence. The solution isn't a single perfect test, but continuously creating new evals that measure capabilities relevant to real-world user needs.

Just as standardized tests fail to capture a student's full potential, AI benchmarks often don't reflect real-world performance. The true value comes from the 'last mile' ingenuity of productization and workflow integration, not just raw model scores, which can be misleading.

Traditional AI benchmarks are seen as increasingly incremental and less interesting. The new frontier for evaluating a model's true capability lies in applied, complex tasks that mimic real-world interaction, such as building in Minecraft (MC Bench) or managing a simulated business (VendingBench), which are more revealing of raw intelligence.